New Blockchain Based Special Keys Security Model With Path Compression Algorithm for Big Data

In recent years, following the introduction of the IoT (Internet of Things) into our lives and thanks to the rapidly increasing number of digital applications, data is collected from a wide variety of sources at an astonishing rate, and the amount of data is increasing exponentially. Today, social networks, cloud computing and data analytics make it possible to collect huge amounts of data. The concept of big data has subsequently emerged and is an important topic in many fields. However, it is not only very difficult to store big data and analyze it, but it is also a serious threat to the security of an individual’s sensitive information. This study describes the issues surrounding big data security and privacy, and provides a solution involving a new blockchain-based security model. This proposed model is called the Blockchain-based Special Key Security Model (BSKM). BSKM proposes, implements and integrates three elements (confidentiality, integrity and availability) of information security together for big data. With this proposed model, a more practical and flexible structure is established for all operations (read, write, update and delete) performed on a database with real data. In this study performed with a special key, all separate blockchain transactions were used for read, write, update and delete operations, and there was a structure that could ensure both confidentiality and integrity at the same time. By looking at a special key for all the blockchain transaction operations performed on the big data it has been shown what type of authorization and access control can be established between which processes and which users. Thus, in contrast to previous studies seen in published literature, data confidentiality, data integrity and data consistency were guaranteed for all transactions. The results of the proposed BSKM model have also been compared by conducting an experimental study of its application. Moreover, this study has shown the effectiveness and benefits of the path compression algorithm. This result has been shown with experimental studies modeling big data and also shows promise for further studies.


23
In recent years, technological advancements have brought 24 about the rise of big data and other digital assets. Big data 25 generates huge amounts of data, and following analysis, 26 making use of this huge amount of information in various 27 scientific and engineering domains. Despite many advantages 28 and applications, there are many challenges involved with big 29 data to be tackled for better quality of service, e.g., analytics, 30 The associate editor coordinating the review of this manuscript and approving it for publication was Jiafeng Xie. management, and privacy and security. Blockchain has the 31 greatest potential to improve big data services and applica-32 tions. The popularity of blockchain technology, and the huge 33 extent of its application, results in much ongoing research in 34 different practical and scientific areas. Blockchain, with its 35 decentralized structure, transparency, auditability and privacy 36 has attracted attention in many sectors with its immutability 37 and the sense of trust it provides. This technology is used in 38 many industries and organizations to improve performance 39 and security. The most important use of blockchain technol-40 ogy is in the cryptocurrency concept and Bitcoin. In addition, 41 (banking and financial data). In addition, the results of the 98 study were carried out on different platforms: Ethereum and 99 Hyperledger. When the path compression algorithm is used, 100 the maximum length and mean length of the chain decreases. 101 Thus, with the path compression algorithm, the long node 102 chain created by the objects is broken, the cost of access-103 ing the objects is reduced, and fast access to the objects is 104 enabled. In short, this study shows that fast access to data can 105 be ensured for big data. 106 The rest of this paper study is organized as follows.: 107 Section section 2 describes the related works regarding stud-108 ies in the published literatüreliterature. Moreover, we discuss 109 the blockchain security solutions that are being used in the 110 area of big data are discussed.; Section section 3 presents 111 big data security and access control issues that need to 112 be solved in order to provide secure data management 113 platforms.; Section section 4, defines the traditional secu-114 rity model.; sIn Section 5, defines the blockchain archi-115 tecture.; Section section 6, describes the study architecture 116 of the proposed system and the its security infrastructure.; 117 Section section 7, presents the experimental results of the 118 proposed model;. Finally, sin Section 8, we presents a per-119 formance evaluation and finally, section 9 presents our the 120 conclusions. As an example, let us consider a scenario that keeps patient 123 records in two medical centers, as in Figure 1. The aim 124 is to share records quickly and securely at both centers. 125 If one center updates the data, the identical patient records 126 should be observed when viewed from either center. In each 127 medical center, a patient's general information is kept, this 128 includes name, surname, TR identity number, date of birth, 129 place of birth, blood group, gender, telephone, address, 130 as well as diagnosis, treatment process, past health findings, 131 VOLUME 10,2022 control. The records of the patients should be shown to the 3) How can the access granted to users be revoked after 157 they complete their transactions? How to prevent access to 158 the system? 159 In short, with the model proposed, the aim is to provide 160 access control and access revocation in a secure manner. 161 In other words, within the scope of this study, the BSKM 162 model that will protect data confidentiality with information 163 flow control for big data is explained. This model was used 164 in a real environment and the effectiveness and success of the 165 algorithm proposed was demonstrated by experimental study. 166 In accessing the data, the cost of access was reduced and the 167 data was accessed quickly. 168 The difference to other studies is that it targets data privacy   At the same time, they can easily change these policies or 185 delete them completely when they are done. This is done 186 during runtime in this study, both statically and dynamically.

187
Unlike other studies, it provides data confidentiality, data 188 integrity and data consistency combined.

190
Due to the development of technology, huge amounts of data 191 are produced in agriculture, finance, banking, business, edu-192 cation, medicine and healthcare. Because diversity in size and 193 format of data is continuously increasing, more flexible data 194 processing tools and platforms are needed to find patterns and 195 useful information in the data [7]. 196 With this large amount of data, information security prob-197 lems have emerged. These problems are data privacy preser-198 vation, identity and access control, data ownership authen-199 tication and authorization. Nowadays, the development in 200 technologies gives rise to concerns about the security and 201 protection of data during storage, transmission, processing 202 and access. Blockchain technology is gaining more attention 203 from the security industry, which is looking for effective ways 204 to secure, protect, store and modify data. Blockchain is a 205 distributed ledger that records transactions linked and secured 206 using cryptography. Transactions can be an exchange of an 207 asset, the execution of the terms of a smart contract, or an 208 update to a record.

209
In recent years, various studies using different techniques 210 for the purposes mentioned have been described in published 211 literature. Lv et al. adopted blockchain technology to solve 212 the privacy protection problem of unmanned aerial vehicle 213 (UAV) big data [5]. In that study, a cryptosystem unit for 214 encryption of the blockchain data was used. This proposed 215 model for analysis using blockchain to protect the privacy of 216 drone big data included a user layer, a data layer, a cloud 217 layer and a blockchain layer. In the user layer, users used 218 blockchain technology to track transactions to prevent shared 219 data from being stolen or tampered with. The information that 220 the user wants to protect is stored in the data layer. The data 221 layer uses a decryption algorithm to recover the original data. 222 The cloud layer is a medium for downloading, uploading, 223 writing and reading data. The blockchain layer can provide 224 a powerful abstraction for distributed protocols. The perfor-225 mance evaluation results show a big data privacy protection 226 scheme based on blockchain technology has low computing 227 costs in terms of key production, encryption and decryption. 228 However, there are some difficulties in implementing this in 229 practice. In addition, this study was only recommended for 230 UAV data. 231 Li et al. proposed a blockchain technique to develop 232 a novel public auditing scheme for integrity recognition 233 in big data in cloud storage [8]. In the proposed scheme 234 that involved three participants, the centralized and expen-235 sive third party auditor was removed, and the computa-236 tion and communication involved were reduced. The aim of 237 that study was to use a blockchain technique for security 238 issues to defend against malicious attacks in cloud stor-239 age. But, more secure and efficient services are required on 240 blockchain-based public auditing techniques. framework for cyber-physical social system big data called 243 BacCPSS [6]. It is important to preserve privacy of access 244 control in CPSS big data by utilizing blockchain fea-245 tures. In BacCPSS, the account address of the node in the 246 blockchain with access control permission is redefined and 247 stored in the blockchain. Also access control and audit in 248 BacCPSS are designed for authorization. Results showed 249 that BacCPSS was feasible and effective, and could achieve 250 secure access in CPSS while protecting privacy. 251 Biswas et al. presented a solution for e-health systems 252 using a unified blockchain-based model [9]. In that study, a     for big data to publish the policies and provide the identifi-295 cation and authentication processes [10]. The policies were 296 visible to the public. Hence, every user could see the policy 297 paired with a resource. However, this needed to use access 298 control tools and a private blockchain. 299 In order to solve the security problem of personnel infor-300 mation management in big data, a personnel management 301 system based on blockchain was proposed [11]. It created 302 a prototype system separating and storing data to solve the 303 problem of blockchain information redundancy and insuf-304 ficient storage space and developed a prototype system to 305 query, add, modify and track personnel information. How-306 ever, this model was not effective in solving such issues 307 as information leakage and tampering. Moreover, the data 308 storage mode proposed by that study needed a better solution 309 to the problem of large-scale data storage in main blockchain 310 technology. 311 Zhang et al. proposed an attribute-based access control 312 scheme that provides decentralized, flexible and fine-grained 313 authorization for IoT devices using blockchain technology 314 [12]. Sultana et al. proposed a blockchain-based smart con-315 tracts data sharing and access control system for commu-316 nication between Internet of Things (IoT) devices [13]. 317 Egala et al. proposed a blockchain-based access model that 318 provided a decentralized and smart contract-based service 319 automation without compromising the system security and 320 privacy. This research introduced hybrid computing with a 321 blockchain-based distributed data storage system for an IoT 322 healthcare system [14]. Zhou et al. proposed a blockchain-323 based access control framework for secured data sharing 324 and smart contract technologies in the Industrial Internet of 325 Things [15]. Lopez et al. presented a blockchain framework 326 for addressing the privacy and security challenges associated 327 with big data in smart mobility. It sent encrypted data to 328 the blockchain network and could make information trans-329 actions with other participants for smart mobility big data 330 [16]. Yang et al. proposed a blockchain-based access control 331 framework with privacy protection called AuthPrivacyChain. 332 They defined the access control permission for data in the 333 cloud, which was encrypted and stored in a blockchain, and 334 they designed processes for access control, authorization and 335 authorization revocation in AuthPrivacyChain [17].

337
The most important problem with big data collected from 338 many areas, such as government agencies, health, education 339 and banking sectors, private enterprises, telecommunications, 340 the Internet, databases of large enterprises, Google, Face-341 book, Yahoo, YouTube and Skype, is security and confi-342 dentiality. With the acquisition of information by users and 343 the collection and sharing of information, security problems 344 related to data confidentiality, data integrity and unauthorized 345 access arise. In this case, the protection of sensitive and 346 valuable personal and public data is mandatory. Therefore, 347 in order to solve the problems arising from the seizure of 348 information by unauthorized persons, information leakage 349 and modification, and non-provision of information confi-350 dentiality and privacy, some studies have been conducted and 351 are seen in published literature [18].  Table 1. Big data security includes all the steps from 403 data collection to data processing, storage, visualization and 404 programming, and it is important at every step.

405
Big data security and privacy challenges are examined in 406 four stages: data collection, data management, system pro-407 gramming and data analysis [20]. Data Collection: Big data consists of real-time structured, 409 unstructured, or semi-structured data obtained from online 410 services, business processes and media. At this stage, con-411 ventional security methods (encryption, authorization, access 412 control, etc.) cannot fully ensure confidentiality and privacy 413 due to the variety of big data (one of the 5V characteristics: 414 velocity, volume, value, variety and veracity), due to the fact 415 that the data consists of different types from different sources. 416 It is not possible to monitor the data traffic that occurs when 417 storing the data because the data increases very quickly. 418 Differences in data format cause security vulnerabilities [21]. 419 Data Management: This stage includes security and confi-420 dentiality issues that occur when storing data collected during 421 the data collection process. The main purpose of data storage 422 is to ensure that authorized individuals in an organization 423 can access the data at any time that they want [22]. But due 424 to the increasing volume of big data, the servers of orga-425 nizations are negatively affected. Conventional data ware-426 houses are not capable of solving this problem. Outsourced 427 data servers such as cloud and distributed systems are used. 428 For example, many necessary or unnecessary data such as 429 customer feedback, e-mails, blogs, social media messages 430 and marketing information can be stored in an e-commerce 431 application. If the processed data is necessary for the organi-432 zation, it is important to analyze and simulate it. Otherwise, 433 too much data is shared unnecessarily over different units 434 within the organization. This, in turn, leads to cost and time 435 losses. In addition, because a large number of operations are 436 performed in storing the same data due to these problems, 437 data integrity is negatively affected. Integrity, which is one 438 of the three basic elements of data security, is not provided 439 by conventional encryption and security methods in multiple 440 structures. It can give attackers the opportunity to hijack 441 servers by taking advantage of these specified vulnerabilities 442 in such structures. Misuse of any data will cause a data leak. securing modern data systems. Institutions use access control 496 models specifically to define who their employees are, what 497 they can do, which resources they can reach, and which 498 processes they can perform. Then they use them to manage 499 the whole process. Access control is a fundamental concept 500 in security by determining who or what can view or use 501 resources in many places and in many businesses. Authen-502 tication and authorization of users and entities is important to 503 minimize the security risk of unauthorized access. The main 504 models of traditional access control are Mandatory access 505 control (MAC), Discretionary access control (DAC), Role-506 based access control (RBAC) and Attribute-based access con-507 trol (ABAC).

508
MAC is a hierarchical model in which access rights are 509 regulated by a central authority by security levels. All users 510 and user groups are allocated a security level. All information 511 is allocated a security label. Users can access or be denied 512 resources that correspond to a security level equal to or lower 513 than theirs in the hierarchy. This system can be quite cumber-514 some to manage because the administrator must allocate all 515 authorization.

516
DAC policy is a means of assigning access rights deter-517 mined by users who have access to their objects. This model is 518 implemented using access control lists by users that can give 519 access authorizations to other users within the limits assigned 520 to them or they can determine the limitations. However, this 521 model increases the risk that data will be made accessible to 522 users that should not necessarily be given access.

523
RBAC provides access rights based on the roles and priv-524 ileges of the users. RBAC requires users to be assigned to 525 different roles to get the associated permissions. However, the 526 problems with role explosion limits its use to enterprise sys-527 tems only. Here, a user may have multiple roles or capacities 528 within a given organization. Thus, when the subject is seeking 529 access to an object, the user must first indicate the role within 530 which the request is being made.

531
ABAC is an authentication and an authorization model 532 that controls access to objects by evaluating rules against the 533 attributes of entities (subject and object). This model requires 534 the basic principles of logical access control. ABAC is an 535 extension of traditional RBAC and can define permissions 536 based on just about any security relevant characteristics, 537 known as attributes. This model is richer and more expres-538 sive because it can be based on any combination of subject, 539 resource and environmental attributes. Today, IoT, cloud computing and social networks make it 542 possible to collect huge amounts of data. Big data has arisen 543 from a growing amount of information that organizations 544 are storing, processing and analyzing [24], [25]. Recently, 545 big data has become a important topic for businesses and 546 government organizations, and in various industry sectors 547 such as healthcare, manufacturing, banking, education and 548 transportation [26], [27]. However, it consists of a variety of 549 big data security and privacy challenges. These challenges 550 are data loss, data breach, data leaking and data theft, and 551 VOLUME 10, 2022 they have become critical threats to organization assets. Tra-552 ditional security access control models are inadequate for 553 solving these problems and are unable to cope with this rapid 554 data explosion. Moreover, these models have failed to cope 555 with the scalability, interoperability and adaptability of big 556 data [28]. Therefore, blockchain-based access controls have 557 been proposed in this current study. with stakeholders inside or outside a company will provide 591 significant improvements for processes such as managing the 592 confidentiality of information. From this point of view, it can 593 be said that the use of this technology in the processes where 594 it is applicable has the potential to deliver great benefits.

596
It is of great importance that the technological infrastructure 597 of the company is compatible with blockchain technology. 598 Investment may be needed in this technology during the 599 adaptation process, and costs such as time and training may 600 need to be considered. However, from a security point of 601 view, data in the blockchain structure can be stored on nodes 602 located in different locations in a distributed manner. If the 603 data is stored openly in an open network structure, there is a 604 risk that the data in these nodes may be intercepted.

605
A transaction is the name given to the process of moving 606 the value of a cryptocurrency from one asset to another in 607 the blockchain network. In this current study, the aim was 608 to ensure data privacy in big data by preventing the capture 609 of valuable data in nodes. The scientific contribution of this 610 study is that data privacy is ensured by defining a private key 611 for each transaction that occurs in the blockchain. This special 612 key has then been applied to real and different large data 613 sets, unlike the studies in published literature. In addition, this 614 private key was created with a bidirectional list data structure. 615 A special path compression algorithm was proposed for data 616 speed while ensuring data security and privacy [29]. This 617 path compression algorithm is implemented for all operations 618 such as adding, deleting, changing and selecting data in the 619 blockchain. The proposed model (BSKM) consists of users, 620 objects and special keys.

622
A user is the person involved in the blockchain transaction. 623 Users include data owners and users, or groups of users, who 624 perform operations such as granting and receiving data autho-625 rization. Each user labels their data for data confidentiality 626 and integrity. The special key consists of a list of security 627 policies that are provided by the users. Each user labels their 628 data for data privacy. That is, a special key is determined that 629 is paired with a data object. In addition, each user has the right 630 to safely change these security policies separately.

632
This refers to the data that users have in the blockchain, that 633 they share with each other, and that they perform various 634 transactions on.

636
A special key is a collection of policies that are created 637 for the protection of data. That is, a key is determined that 638 is paired with a data object. This special key contains the 639 encrypted version of the data sent, added, deleted or updated 640 for each transaction that takes place in the blockchain. The 641 special key is encrypted by creating policies. In addition, each 642 user has the right to safely change these security policies 643 separately. This model was developed for unreliable users and 644       Here, while 'owner' denotes the actors who own the labeled 680 object, 'readers' refers to the users to whom authorization is 681 given to read the owners' data transactions; 'writers' refers to 682 the actors to whom authorization is given to write to the data 683 owners' transactions; 'updaters' refers to the actors to whom 684 authorization is given to update the data owners' transactions 685 and 'deleters' refers to the users to whom authorization is 686 given to delete data owners' transactions. The example spe-687 cial key shown with the doubly linked list can be expressed 688 in the S typing format as follows [27]: The semicolon used when creating a label separates the 692 policies from one another. Accordingly, the S label has five 693 policies: {U 1 :U 2 ,U 4 }, {U 2 :U 3 , U 4 }, {U 3 :U 4 , U 5 }, { U 4 :U 5 } 694 and {U 5 : }. While U 1 , U 2 , U 3 , and U 4 denote the owners of 695 the data object to which the Slabel belongs, U 2 , U 3 , U 4 and v 5 696 represent the actors authorized by the data owners for various 697 object transactions (read, write, update and delete). 698 Let us assume that the first policy shows the read operation 699 on the object:

700
The first policy is expressed by U 1 → U 1 , U 1 → U 2 and 701 U 1 → U 4 edges. This means that the U 1 user allows the v 1 , 702 v 2 and v 4 users to read their data. 703 Let us assume that the second policy shows the write 704 operation on the object:

705
The second policy is expressed by U 2 → U 2 , U 2 → U 3 , 706 and U 2 → U 4 edges. This means that the v 2 user allows the 707 U 2 , U 3 and U 4 actors to write to their data. 708 Let us assume that the third policy shows the update oper-709 ation on the object:

710
The third policy is expressed by U 3 → U 3 , U 3 → U 4 , 711 U 3 → U 5 edges. This means that the U 3 user allows the U 3 , 712 U 4 and U 5 actors to read their data. 713 Let us assume that the fourth policy shows the delete 714 operation on the object: 715 This is expressed by U 4 → U 4 , U 4 → U 5 edges. This 716 means that the U 4 user allows the U 4 and U 5 users to delete 717 their data.

718
The last policy is expressed by the U 5 → U 5 edge. This 719 means that U 5 does not allow anyone other than themselves 720 to perform any transaction on their data.

722
When a user performs any blockchain transaction (read, 723 write, update, delete), they may want to revoke this access 724 right. In this case, private keys are determined for the object 725 again and the access right granted is revoked. With the 726 re-issued private key, the data is made less restrictive and less 727 stringent.

728
With the user hierarchy in Figure 5, client_X user allows 729 the lawyers group to read their legal data. 730 Figure 6 shows the user hierarchy in a law firm. The 731 lawyers user creates a group and all users in this group 732 (lawyer_X, lawyer_Y, lawyer_Z) are members of this group. 733 These lawyers exercise the authority and responsibility for 734 this group. Considering the user hierarchy, the action to re-use 735 4 different private keys regarding the revocation of the access 736 right is given below: 737 VOLUME 10, 2022   the lawyer_X: lawyer_Z addition needs to be cancelled, the 763 client_X can re-determine the private key of the legal data 764 {client_X: lawyer_X} with the private key. By deleting a 765 new policy from a private key, data becomes more tightly 766 protected.

767
Action 4: Self-authorization. Let S be the private key 768 {X:Y}. X is the data owner and Y is the reader of this tag. 769 Also, let the user X give the power of attorney to users Z and 770 T. Users given the power of lawyers are added to the data as 771 a reader with their private key {X:Y}. Thus, the data is more 772 restricted with the newly created private key {X:Y,Z,T}. At a 773 law firm, the lawyers authorized by the client can represent 774 the client by adding them as readers to the new label. With 775 the {client_X: lawyer_X} tag, client_X authorizes lawyer_X 776 to conduct their own case. Considering the user hierarchy in 777 Figure 5 and Figure 6, client_X gives the power of lawyers 778 to all lawyers in the lawyers group. By adding lawyer_Y 779 and lawyer_Z as readers to the old private key, the data 780 becomes more restricted with the newly created private key 781 {client_X:lawyer_X, lawyer_Y, lawyer_Z}.

782
Any transaction can be canceled by the re-switching rules 783 given above because these rules are recreated without break-784 ing the security level. For example, in the multi-level secu-785 rity model, each user includes their own data in one of 786 these classes, including unclassified, classified, secret and top 787 secret, in order to protect their own data, and private keys are 788 determined again. Each user in each security class acts on 789 behalf of its own class, and these users can see all transactions 790 in its class.

791
The transactions for revoking the access rights granted by 792 a data object from user ui to user uj are shown below in 793 pseudocode. Meanwhile, in order for the user uj to revoke the 794 access right (read, write, update, delete) granted to the user ui 795 -a policy in S -which is the private key for this data, it must 796 be included in the data owner or all authorized lists. This is 797 expressed by the following condition:  The two real datasets used in the study taken from different 843 sectors underwent a preliminary process so that each user 844 and object mentioned in the dataset was classified according 845 to the security transactions. Real classification scales for 846 these institutions were taken as the basis for the classification 847 process. Datasets were labeled as the bank dataset and the 848 financial dataset.

849
The first dataset is a real dataset obtained from a Turk-850 ish bank. Moreover, this dataset consisted of approximately 851 10 million customers and nearly 100 million transactions 852 such as money transfers and electronic fund transfers. This 853 dataset contained confidential and sensitive information. For 854 this reason, no information is given about the content of the 855 data.

856
The financial data is what investors use to analyze a 857 company's economic and financial health. It contained a 858 100 million different transactions. This dataset also contained 859 confidential and sensitive information. For this reason, it was 860 not shared between different environments.  The time performance criterion was calculated by Equa-880 tion3:

881
Time= System uptime for a transaction (3) 882 In Table 2 and Table 3, the success of the proposed model 883 was compared to published literature methods in terms of 884 VOLUME 10, 2022   accuracy against real data sets, which has been taken respec-885 tively from the data of a bank and a financial institution whose 886 classes are obvious. Accuracy rates were calculated using 887 transactions randomly selected from these data sets. In addi-  The proposed model I was compared with some datasets 908 used in published literature in Table 4    bank data and in Figure 13 for financial data. This study 925 built an Etherum blockchain platform, firstly by downloading 926 and installing the latest version of Ethereum (geth1.7.3 [11]) 927 on four servers in the same local area network, deploying 928 four separate nodes, then specifying the same network id 929 and genesis.json for all nodes to ensure that nodes could 930 be properly connected to each other, finally completing the 931 connection between the main node and child node by file 932 configuration.

933
Hyperledger is an open source project created to sup-934 port the development of blockchain-based distributed ledgers. 935 Hyperledger consists of a collaborative effort to create the 936 necessary frameworks, standards, tools and libraries to build 937 blockchains and related applications. This study built a 938 blockchain platform of Hyperledger Fabric by firstly down-939 loading and installing the Hyperledger Fabric v0.6 on the 940 server, then configuring the fabric environment, and finally 941 deploying the fabric network and testing the chaincode. The 942 results showed that the maximum number of nodes that Fabric 943 v0.6 can have is 40 nodes in the network.

944
The results of the model for bank data on the Etherum 945 platform are shown in Figure 14. Separately, all transactions 946 for financial data on the Hyperledger platform are shown in 947 Figure 15. When the proposed model was compared with 948 traditional security models, it was observed to give more 949 successful results on both platforms.  Table 6 the success of the proposed model (BSKM) is 952 compared with published literature in terms of time against 953 the real data set taken from the bank. In Table 7, the success 954 of the proposed model (BSKM) is compared with published 955 literature in terms of time against the real data set taken from 956 the financial data. Accuracy rates were calculated by random 957 VOLUME 10, 2022 FIGURE 14. Traditional security models and BSKM accuracy rate for bank data in Hyperledger platform.  selections from these data sets. The success of the proposed 958 model is clearly shown in Figure 16 and in Figure 17 for the 959 bank and the financial data, respectively. In terms of time, 960 it was seen that operations were performed on the data in  time than other transactions. This is because it takes more 971 time to write and update the object.

973
The developed model being proposed in this study has been 974 applied on a dataset obtained from real life bank data. 975 The performance of the proposed model has been com-976 pared with the performances of traditional access control 977 models. When the results obtained were compared, it was 978 observed that object access levels were presented more con-979 sistently and quickly with the proposed model. The accuracy 980 results obtained by using the path shortening algorithm with 981 the BSKM model and the traditional recommended access 982  control models are shown in Table 8. Looking at Figure 18 Table 9 shows the comparison of the proposed BSKM    algorithm, its effectiveness was also demonstrated by the 996 experimental study.

998
In this study, BSKM was introduced to ensure big data secu-999 rity. In the proposed model, both authorized and unautho-1000 rized operations between users were carried out. Also, in the 1001 proposed model, there was separate authorization or access 1002 control for each blockchain transaction, such as reading, 1003 writing, updating and deleting. Access control and authoriza-1004 tion operations were performed using special keys. Unlike 1005 previous studies, data security was ensured for all operations 1006 performed on the big data. Users could take back the authority 1007 that they gave at any time, or they can give authority to the 1008 user they want. Challenges that occurred during the imple-1009 mentation of security policies on big data were overcome.